lab 2 · data frames (R) — source
A. Stamm 
This page shows the source file of the lab below. Use the button to download it as a plain file and run it in your own workspace.
# ============================================================
# Lab 2: Data frames, tibbles & reading data
# The R Statistical Language · Session 1
#
# The goal: load a "messy" penguins file, diagnose its shape,
# fix the column types, and write a clean CSV. This is your
# first taste of the data-cleaning decisions you'll formalise
# in the Reproducible Workflows course.
# ============================================================
## Part A: build and explore a data frame -------------------
# A1. Create df <- data.frame(
# x = 1:5,
# y = c("a","b","c","d","e"),
# z = c(TRUE, FALSE, TRUE, TRUE, FALSE)
# )
# Inspect it with str(df), summary(df), head(df).
# What is class(df)? What is the class of df$x? of df$y?
# A2. Convert to a tibble: df_t <- tibble::as_tibble(df)
# Print it. How does the print differ from a data.frame?
# A3. Subset:
# - rows 2 and 3 of column 1;
# - all rows of columns 1 and 3;
# - all rows where z is TRUE.
## Part B: read the messy file ------------------------------
# B1. Open penguins_messy.txt in a text editor (outside R).
# Observe: what separates the values? Is there a header?
# Are there missing values, and what do they look like?
# B2. Try reading with:
# library(readr)
# read_csv("penguins_messy.txt") # -> probably wrong!
# read_table("penguins_messy.txt") # -> probably closer
# Look at the "spec" (column types) readr guessed via
# problems() or the printed column spec. Which columns are
# wrong (should be numeric but came in as character)?
# B3. Read it successfully and fix the types:
# penguins <- readr::read_table(...)
# for columns that should be numeric, coerce, e.g.
# species, island and sex should be factors (or kept character).
## Part C: cleaning decisions -------------------------------
# C1. Missing values: how many NAs per column?
# colSums(is.na(penguins)) (is.na -> logical; colSums sums TRUEs)
# C2. Are the type coercions what you expect? Use str() to confirm.
# C3. Write the cleaned data:
# write_csv(penguins, "penguins_clean.csv")
# Re-open it with read_csv(): are the types still right
# after the round trip?
## Part D: STRETCH -------------------------------------------
# D1. Why did read_csv() mis-guess column types on this file?
# Experiment with a smaller sample (n_max = 3) to diagnose.
# D2. For the island column, use table(island) to summarise counts.
# Turn island into a factor with forcats::fct_infreq(island)
# and re-tabulate: what changed, and why might that order help
# a later barplot?
# D3. The {janitor} package (used in RR course) has clean_names().
# Install it (install.packages("janitor")) and run
# janitor::clean_names() on penguins: what does it change?
# (This foreshadows the data-cleaning toolchain.)